Abstract
In the first chapter of this dissertation, I estimate the welfare gains from rural roads in Ethiopia. I develop a Ricardian trade model to explain how rural roads affect welfare in agricultural villages. The model captures two contrasting effects of rural roads on welfare. On one hand, decreases in trade costs lead to reallocation of village farmland to its comparative advantage crops and enable farmers to obtain higher prices for these crops. On the other hand, as these crops constitute significant part of the household consumption basket, the increase in prices decreases welfare. I take these predictions of the model to a very rich agricultural data. I show that, following decreases in trade costs due to massive rural road expansion, villages reallocate more farmland toward their comparative advantage crops and receive better prices for these crops, thus receiving higher nominal revenue. Despite increases in village price index, the gain in nominal income dominates, resulting in net welfare gain, on average. The size of this welfare gain depends on the crop composition of village consumption vis-a-vis production.
In the second chapter, I suggest a new approach to test separability-- whether farm household's production and consumption decisions can be separately analyzed -- and explore its link to market integration. The existing separability tests looked at the link between on-farm labor demand and household demographic characteristics. One problem with this approach is that on-farm labor demand is likely to be poorly measured in the context of self-employing agricultural households. I suggest an alternative test that is based on the link between household land allocation across different crops and the household's consumption tastes. I first estimate households' tastes for crops from their preference functions, and show that these tastes significantly dictate the households' land allocation across crops. The extent to which crop tastes dictate land allocation decreases with improvement in market integration due to construction of new rural roads.
In the third chapter, I study how frequent booms and busts in international coffee prices affect welfare of coffee producers in Ethiopia. I use unique panel data on household consumption and land utilization to show that decreases in international coffee prices are passed on fully to consumption expenditure in households that allocate all farmland to coffee production. The decline in consumption has real consequences on child health and mortality. I find that children born in coffee producing districts during low coffee price periods are more likely to be underweight, wasted, stunted, and anemic, compared to their peers born in non-coffee districts; and their under-five mortality rate is higher by 1.8 percentage points.
